Control design of step-down type DC-DC converter against sudden change in battery voltage for electric vehicles

This thesis presents a novel gain-scheduled feedforward control strategy tailored for DC-DC buck converters subjected to abrupt input voltage variations. The design is based on a quasi-LPV model constructed using the PWM duty cycle as the scheduling parameter, where parasitic resistances are explicitly included to enhance modeling accuracy. An affine LPV compensator is synthesized via an H∞ optimization procedure, leveraging a dilated Bounded Real Lemma (BRL) formulation that incorporates a parameter-dependent Lyapunov function. The proposed compensator is integrated into a modular architecture that allows compatibility not only with classical PI controllers but also with other feedback-based control schemes. Numerical simulations and experimental validations confirm that the control approach effectively attenuates output voltage fluctuations and significantly reduces current ripple. For example, during rapid 48 V ↔️ 19.5 V input transitions, output voltage deviations are constrained within ±0.1 V, achieving a performance improvement of up to 98.5%. Beyond buck converters, the proposed methodology is extendable to general LPV systems exhibiting parameter-varying dynamics. With its low computational complexity and robustness, the framework is well-suited for real-time power electronics applications requiring high-performance regulation and control adaptability.